The Aleutian Tsunami of 1946: the Compound Earthquake-Landslide Source and Near-Field Modeling

Abstract:

The tsunami of April 1, 1946, spread death and destruction throughout the Pacific from the Aleutians to Antarctica, and produced exceptional runup, 42 m, at Scotch Cap on Unimak Island in the near field. López & Okal (2006) showed that the triggering earthquake was at least M_W = 8.6, large enough to explain the far-field tsunami but still requiring a landslide or other secondary source to achieve the local runup. No convincing landslide was found until von Huene, et al (2014) merged all available multibeam data and reprocessed a old multichannel line to show that a feature on the Aleutian Terrace they call Lone Knoll (LK) is the displaced block of a translational slide. From ²¹⁰Pb dating of push cores taken near the summit of LK, we find that a disruption in sedimentation occurred in 1946 at one site, but sedimentation was not disrupted at another site nearby. We infer that the slide block moved coherently at a speed close to the threshold for erosion of the hemipelagic clays. From GLORIA sidescan, Fryer, et al (2004) had earlier tentatively identified LK as a landslide deposit, but if the tsunami crossed the shallow Aleutian Shelf at the long-wave speed, that landslide had to extend up to the shelf edge to satisfy the known 48-min travel time to Scotch Cap. The resulting landslide was enormous, and a multibeam survey later in 2004 showed that it could not exist. The slide imaged by von Huene, et al is far smaller, with a headwall 30 km downslope at a depth of 3 km. The greater distance demands that the tsunami travel much faster across the shelf. The huge runup, however, suggests that wave height was a significant fraction of the water depth (only 80 m), so the tsunami probably crossed the Aleutian Shelf as a bore. From modeling the landslide-generated tsunami with a shock-capturing dispersive code we infer that it did indeed cross the shelf as a bore traveling at roughly twice the long-wave speed. We are still exploring the dependence of the tsunami on slide parameters, but our combined earthquake-slide model broadly matches the near-field runup and the observed arrival time at Scotch Cap.